贝叶斯时空模型在新冠风险评估中的应用

Application of Bayesian Spatio-TemporalModel in the Risk Assessment of COVID-19

摘要: 基于2020~2021年河南省18个地级市新冠肺炎确诊病例数据，采用BYM (Besag-York-Mollie)、FBM (Familiar Bayesian Spatio-temporal)模型来分析新冠肺炎发病风险的时空分布特征，并探讨社会经济、 生态环境等因素对疾病风险的影响效应，同时就应用结果对比两种模型的优劣。方法：收集2020~2021 年河南新冠肺炎确诊数据，采用相关性分析排除部分相关性不大的影响因子，并运用因子分析降维诸指 标，而后分别代入两模型进行分析、讨论。结果：1) 2020~2021年河南全省的疫情风险程度是呈现下 降趋势的，高风险地区主要集中于豫北地区。2) 因子分析共构建基本经济因子、社会因子、温度因子、 降水因子以及空气质量因子五个综合指标。3) 通过模型运行后的结果和DIC值可以得出FBM模型总体要 优于BYM模型。

Abstract: Based on the data of confirmed cases of COVID-19 in 18 prefecture-level cities in Henan Province from 2020~2021, BYM (Besag-York-Mollie) and FBM (Familiar Bayesian Spatio-temporal) models were used to analyze the spatial and temporal distribution characteristics of the risk of COVID-19,and explore the influence of socioeconomic and ecological factors. The results of this study were also used to compare the advantages and disadvantages of the two models. Methods: The data of confirmed cases of COVID-19 in Henan from 2020 to 2021 were collected, correlation analysis was used to exclude some influencing factors with little correlation, and factor analysis was applied to downscale the indicators, which were then analyzed and discussed by substituting into the two models. Results: 1) The risk level of the epidemic in Henan province from 2020 to 2021 showed a decreasing trend, and the high-risk areas were mainly concentrated in the northern part of Henan. 2) The factor analysis constructed five comprehensive indicators: basic economic factor, social factor, temperature factor, precipitation factor and air quality factor. 3) The results and DIC values after the model run can be concluded that the FBM model is better than the BYM model in general.